作者： Jakub Rusinowski · 最后更新： 2026年7月30日
These are the strongest local models that fit entirely in 10 GB of VRAM, ranked by capability, with the quantization level and estimated tokens/sec needed to fit.
| Qwen3-Coder — Qwen3-Coder 80B-A3B (MoE) | Q4_K_M · 7.5 GB · ~400 tok/s on Intel Arc B570 |
| Qwen 3 — Qwen 3 30B-A3B (MoE) | Q4_K_M · 8 GB · ~400 tok/s on Intel Arc B570 |
| Granite 3.0 — Granite 3.0 8B Instruct | Q4_K_M · 5.5 GB · ~69 tok/s on Intel Arc B570 |
| Bonsai 27B — Ternary Bonsai 27B | Ternary (1.58-bit, ~1.71 bpw) · 5.9 GB · ~64 tok/s on Intel Arc B570 |
| Qwen 2.5 Family — Qwen 2.5 7B Instruct | Q4_K_M · 4.8 GB · ~79 tok/s on Intel Arc B570 |
| Qwen 3 — Qwen 3 8B | Q4_K_M · 5.5 GB · ~69 tok/s on Intel Arc B570 |
| DeepSeek R1 — DeepSeek R1 Distill Llama 8B | Q4_K_M · 5.8 GB · ~66 tok/s on Intel Arc B570 |
| Qwen3-Coder — Qwen3-Coder 8B | Q4_K_M · 5.5 GB · ~69 tok/s on Intel Arc B570 |
| Mistral Family — Mistral NeMo 12B | Q4_K_M · 8.5 GB · ~45 tok/s on Intel Arc B570 |
| Gemma 4 (Legacy Listing — Unverified) — Gemma 4 12B | Q4_K_M · 7.8 GB · ~49 tok/s on Intel Arc B570 |
| Gemma 3 — Gemma 3 12B Instruct | Q4_K_M · 8.1 GB · ~47 tok/s on Intel Arc B570 |
| Qwen 3.5 (Legacy Listing — Unverified) — Qwen 3.5 7B | Q4_K_M · 4.8 GB · ~79 tok/s on Intel Arc B570 |
| GLM-6 — GLM-6 9B | Q4_K_M · 6 GB · ~63 tok/s on Intel Arc B570 |
| InternLM 3 — InternLM 3 8B Instruct | Q4_K_M · 5.5 GB · ~69 tok/s on Intel Arc B570 |
| Qwen 3.5 — Qwen 3.5 9B | Q4_K_M · 6.6 GB · ~58 tok/s on Intel Arc B570 |
Qwen3-Coder, Qwen 3, Granite 3.0, Bonsai 27B, Qwen 2.5 Family all fit in 10 GB VRAM.
Intel Arc B570, NVIDIA GeForce RTX 3080 (10GB), Intel Arc B580, NVIDIA GeForce RTX 3060 (12GB).